I can't f*cking believe Sam Altman said this out loud.
He's "probably not supposed to say this": OpenAI's new model crushed researchers' hardest problems one after another, and it made him sad:
"The model's just knocking down one after the other... I also had much more sadness and melancholy and a feeling of something ending than I was expecting"
If the person building it feels that, don't wait. Start here:
Anthropic Academy: free courses from the people who built Claude
anthropic.skilljar.com
Anthropic Courses: prompting and tool use, from the source
github.com/anthropics/course…
Microsoft - Generative AI for Beginners: 21 lessons, zero to building
github.com/microsoft/generat…
Microsoft - AI Agents for Beginners: AI that works on problems while you sleep
github.com/microsoft/ai-agen…
Hugging Face - Agents Course: build and ship your first agent
github.com/huggingface/agent…
Building effective agents (Anthropic): what actually works when you hand AI hard problems
anthropic.com/research/build…
How Anthropic built its multi-agent research system: many models attacking one problem
anthropic.com/research/multi…
OpenAI Cookbook: OpenAI's own working examples
github.com/openai/openai-coo…
Karpathy's autoresearch: an agent that runs research loops on its own
github.com/karpathy/autorese…
Andrej Karpathy - Neural Networks: Zero to Hero: understand the model, not just use it
github.com/karpathy/nn-zero-…
DAIR.AI - Prompt Engineering Guide: every technique in one place
github.com/dair-ai/Prompt-En…
What's a mob to a king, what's a king to a god
95% will feel the chapter closing and do nothing
5% will be the ones handing the model their hardest problems
Save this and send it to someone who still thinks they have years to figure AI out
Then read the full guide below ↓
14
1,013
Hercyles retweeted
This is fucking insane.
A SpaceXAI engineer just described the choice every engineering manager now faces. Hire a person, or buy tokens:
In 82 seconds Lauren Tan explains why a startup doesn't want to "become a 10,000-person engineering org"
"It would have taken me years to build this framework... And my salary is pretty high"
Once the codebase is set up right, "even agents that are not, you know, fable size, do an excellent job of writing code"
Hire An Engineer → Pay The Salary → Spend Tokens Instead → Set Up The Codebase → The Engineer Who Builds The Setup
Everything below is free. Start today:
Claude Code best practices: how the people who built it actually use it
code.claude.com/docs/en/best…
Claude Code - CLAUDE.md + memory: the rules every agent reads before it touches your code
code.claude.com/docs/en/memo…
Claude Code - Skills: turn what you know into instructions any agent can follow
code.claude.com/docs/en/skil…
Claude Code - Hooks: lint and tests run on every change, no exceptions
code.claude.com/docs/en/hook…
Anthropic Skills: ready-made skills to start from
github.com/anthropics/skills
Building effective agents (Anthropic): the patterns that hold up in real codebases
anthropic.com/research/build…
Context engineering for agents (Anthropic): why the same model does great work in one repo and garbage in another
anthropic.com/research/effec…
Anthropic Cookbook: working code for agents, tools and evals
github.com/anthropics/claude…
12-Factor Agents: engineering rules for agents you can trust in production
github.com/humanlayer/12-fac…
GitHub Spec Kit: write the spec once, let agents build against it
github.com/github/spec-kit
Superpowers: skills that make coding agents plan, test and verify
github.com/obra/superpowers
OpenAI Agents SDK: OpenAI's own framework for multi-agent work
github.com/openai/openai-age…
MCP servers: connect agents to your repo, docs and databases
github.com/modelcontextproto…
Save this and send it to an engineer who still thinks the job is writing code
This is fucking insane.
A SpaceXAI engineer just described the choice every engineering manager now faces. Hire a person, or buy tokens:
In 82 seconds Lauren Tan explains why a startup doesn't want to "become a 10,000-person engineering org"
"It would have taken me years to build this framework... And my salary is pretty high"
Once the codebase is set up right, "even agents that are not, you know, fable size, do an excellent job of writing code"
Hire An Engineer → Pay The Salary → Spend Tokens Instead → Set Up The Codebase → The Engineer Who Builds The Setup
Everything below is free. Start today:
Claude Code best practices: how the people who built it actually use it
code.claude.com/docs/en/best…
Claude Code - CLAUDE.md + memory: the rules every agent reads before it touches your code
code.claude.com/docs/en/memo…
Claude Code - Skills: turn what you know into instructions any agent can follow
code.claude.com/docs/en/skil…
Claude Code - Hooks: lint and tests run on every change, no exceptions
code.claude.com/docs/en/hook…
Anthropic Skills: ready-made skills to start from
github.com/anthropics/skills
Building effective agents (Anthropic): the patterns that hold up in real codebases
anthropic.com/research/build…
Context engineering for agents (Anthropic): why the same model does great work in one repo and garbage in another
anthropic.com/research/effec…
Anthropic Cookbook: working code for agents, tools and evals
github.com/anthropics/claude…
12-Factor Agents: engineering rules for agents you can trust in production
github.com/humanlayer/12-fac…
GitHub Spec Kit: write the spec once, let agents build against it
github.com/github/spec-kit
Superpowers: skills that make coding agents plan, test and verify
github.com/obra/superpowers
OpenAI Agents SDK: OpenAI's own framework for multi-agent work
github.com/openai/openai-age…
MCP servers: connect agents to your repo, docs and databases
github.com/modelcontextproto…
Save this and send it to an engineer who still thinks the job is writing code
This is fucking insane.
Sam Altman was asked how he finds the bets nobody else sees. His answer is a skill school never taught him:
"Very little of my experience in school was figuring out what questions to ask
I think we should focus much more on teaching very young people to figure out what questions to ask"
In 70 seconds he adds the career trap most people never notice:
"Sure you have 20 years of experience, but you have 20 times the same one year of experience"
His fix: "keep doing uncomfortable, non-obvious new things". Staying on that curve, he says, "is empirically not what most people do"
School Answers → Same Year On Repeat → AI Gives The Answers → New Questions → The Person Who Knows What To Ask
AI is now the cheapest answer machine ever built. What it can't do is decide what's worth asking. That's the skill to train, and every tool you learn adds a new year instead of repeating the old one
Everything below is free. Start today:
Anthropic Academy: free courses from the team behind Claude
anthropic.skilljar.com
Elements of AI: the no-math intro to how AI works
elementsofai.com
Harvard CS50 - Intro to AI with Python: Harvard's own course, $0
cs50.harvard.edu/ai/
Google - Machine Learning Crash Course: how Google trains its own engineers
developers.google.com/machin…
Microsoft - AI for Beginners: 12 weeks, 24 lessons
github.com/microsoft/AI-For-…
fast.ai - Practical Deep Learning: build first, theory later
course.fast.ai
Kaggle Learn: short hands-on lessons, right in the browser
kaggle.com/learn
Hugging Face Learn: every course from the home of open models
huggingface.co/learn
DeepLearning.AI - Courses: short courses with the people who build the models
deeplearning.ai/courses
Andrej Karpathy - Neural Networks: Zero to Hero: what's under the hood
github.com/karpathy/nn-zero-…
Anthropic - Prompt Engineering Interactive Tutorial: learn to ask the model better questions
github.com/anthropics/prompt…
Learn Prompting: the free prompting course for beginners
learnprompting.org
Total cost: $0
95% will repeat the same year of experience again in 2027
The 5% who spend it on something uncomfortable will be the ones Altman is describing
Save this and send it to a parent whose kid is still in school
Then read the full guide below ↓
this is f*cking insane.
Sam Altman told Stanford MBA students that whole categories of jobs will be done entirely by AI. Then he said this is exactly when you want to start a career:
"Wide swaths of current jobs, the way they're done today at least, will be entirely done by AI
People with drive and ideas and new skills do relatively the best when the ground is shaking the most"
In 85 seconds he walks through what already happened to programmers: two years of "our job is over", then "we're going to keep paying ever higher salaries to people that can get computers to do things"
His verdict: no "job apocalypse", but "huge amounts of short-term changeover"
Old Job → AI Takes The Tasks → New Skills → Agents → The Person Who Gets Computers To Do Things
The salary moves to whoever can get AI to do the work. That's a skill, and every piece of it is free to learn
Everything below is free. Start today:
Anthropic Academy: free courses from the team behind Claude
anthropic.skilljar.com
Claude Code best practices: from the people who built it
code.claude.com/docs/en/best…
OpenAI Cookbook: code for every common task
github.com/openai/openai-coo…
OpenAI Agents SDK: a team of agents in a few lines
github.com/openai/openai-age…
Microsoft - Generative AI for Beginners: 21 lessons, no degree required
github.com/microsoft/generat…
Hugging Face - Agents Course: free and certified, start to finish
github.com/huggingface/agent…
DAIR.AI - Prompt Engineering Guide: the reference everyone links to
github.com/dair-ai/Prompt-En…
Andrej Karpathy - Neural Networks: Zero to Hero: what's under the hood
github.com/karpathy/nn-zero-…
GitHub Spec Kit: write the spec, let the agent build it
github.com/github/spec-kit
Superpowers: skills that make coding agents plan, test and check their work
github.com/obra/superpowers
opencode: open-source coding agent for the terminal
github.com/anomalyco/opencod…
OpenHands: an AI developer that writes and fixes code by itself
github.com/OpenHands/OpenHan…
MarkItDown: turn PDFs, Word and Excel into text your AI can read
github.com/microsoft/markitd…
Total cost: $0
95% will spend this year arguing about whether AI takes jobs
The 5% who learn to get computers to do things will collect the higher salaries Altman is talking about
Save this and send it to someone starting their career
Better then 3250$ AI course
2
22
2,617
This is f*cking insane
Nvidia ex-CEO Jensen Huang got asked what he'd say to everyone scared AI will take their job. His answer:
"You're going to lose your job to somebody who learned AI better than you"
In less than 2 minutes, he breaks it down: the PC never took anyone's job. It left behind the people who refused to learn it
AI is everywhere inside Nvidia, and the company is still hiring. Stuff he expected in 10 years, he now expects in 1 or 2
Learn AI → Use AI → Build Agents → Harness → The One Nobody Replaces
Save this before your feed buries it
Every link below is 100% free:
Elements of AI, a free intro with zero coding
elementsofai.com
fastai's Practical Deep Learning course
course.fast.ai
Google's Machine Learning Crash Course
developers.google.com/machin…
21 lessons on building with generative AI
github.com/microsoft/generat…
Hugging Face's free AI agents course
github.com/huggingface/agent…
Build a ChatGPT-style model from scratch
github.com/rasbt/LLMs-from-s…
Run any AI model right from your terminal
github.com/simonw/llm
Turn PDFs, Word and Excel into text your AI can read
github.com/microsoft/markitd…
Plug-in tools for your agent
github.com/modelcontextproto…
Skills that make coding agents plan, test and check their own work
github.com/obra/superpowers
An AI developer that writes and fixes code by itself
github.com/OpenHands/OpenHan…
Open-source coding agent for the terminal
github.com/anomalyco/opencod…
A local agent that installs, runs and edits for you
github.com/aaif-goose/goose
Total cost: $0
AI won't take your job. The person who learned it first will
Then build your harness with the guide below before someone else learns AI better than you ↓
1
10
583
Former Google CEO Eric Schmidt just told everyone the easiest way to get rich in the next two years, and it's not stocks or crypto:
"If you really want to make money, it's actually easy, found an agentic AI company, I mean build an agent to do something
If you build a better agent than anyone else, you can probably build one incredible company and own it yourself"
The man who ran Google for 10 years explains in under 30 seconds why the next year or two is the agentic period, and why that window won't stay open for long
Prompts → Claude Code → Loops → Graphs → Gates → A Company You Own Yourself
Save this before it slips out of your feed
Everything below is free:
Building effective agents (Anthropic)
anthropic.com/engineering/bu…
How Anthropic built its multi-agent research system
anthropic.com/engineering/mu…
Harnesses for long-running agents
anthropic.com/engineering/ef…
Writing tools for agents
anthropic.com/engineering/wr…
Claude Code 101
anthropic.skilljar.com/claud…
Claude Code in action
anthropic.skilljar.com/claud…
Prompt engineering
platform.claude.com/docs/en/…
Subagents (parallel work)
code.claude.com/docs/en/sub-…
Hooks (checks that can say no)
code.claude.com/docs/en/hook…
CLAUDE.md + memory
code.claude.com/docs/en/clau…
Skills
code.claude.com/docs/en/skil…
MCP
code.claude.com/docs/en/mcp
LangGraph (graphs of agents)
langchain-ai.github.io/langg…
Karpathy's autoresearch (an agent in a loop)
github.com/karpathy/autorese…
All of it costs $0
95% will spend the agentic period scrolling AI threads. The 5% who ship agents will own the companies that come out of it
Bookmark it
Then read how to build an agent that works for hours without you in the Loops & Graphs guide below ↓
15
644
This is f*cking insane. Elon Musk put SpaceXAI's best engineers on a live stream and gave them 72 hours:
"Empty GitHub org. Zero bots. Build a whole company with Grok Bot, live"
In 52 minutes, they hire the first bots, build one from a single voice prompt, teach it a task by doing it once and set a 9 am routine
One Bot Per Job → Voice Prompt → Approval Rules → Skills → Daily Routines → A Company Of Bots
Nobody writes these bots by hand. They describe the job out loud, show it once, and the bot takes over from there
95% will keep opening a fresh chat for every task. The 5% who watch this will have bots doing it before they wake up
Better than any $5,000 AI automation course
Then the Grok Bot guide below shows how to point that same company of bots at the markets ↓
1
12
323
This is f*cking insane. Anthropic Research Lead:
"Prompting is basically over 99% of our engineers already run swarms of 300+ self-improving agents. Now everyone is shifting to graphs"
In just 20 minutes, she explains how graph engineering
turns separate agents into a system that improves itself
One Agent → Loops → Plan Mode → Graphs → Self-Improving System
Prompting was the old workflow.
The graph around the model is the new one
Worth more than most $1,000 agent engineering courses
Bookmark and watch it today
Then read the article below ↓
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
1
8
438
Anthropic's president says she learned to run the company from playing the flute.
Daniela Amodei:
"On the other side is this world in which human skills are the main thing that's left"
58 minutes with Arianna Huffington on Claude, jobs and human skills.
Watch it this weekend
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
11
256
The man who backed Musk, Altman and Zuckerberg says the real AI danger isn't the AI.
Peter Thiel:
"We need to hit the accelerator as hard as possible"
58 minutes on Anthropic as a $2T "cult", Armageddon vs the Antichrist and why evil beats bad.
Watch it before Monday
One solo founder pays Perplexity $1M a year, because it makes him $3M Aravind Srinivas:
"AI is a recursive product"
55 minutes on Perplexity Computer, running agents on your own laptop and why he has no competitors, only contributors.
Bookmark it before your next pitch
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
1
16
624
Demis Hassabis watched his AI go from random moves to beating every chess engine in one day
Now he's worried:
"Closed-loop RSI is a very dangerous way to get to full AGI"
25 minutes on AlphaFold, AI consciousness and why we're not ready.
Save it before your next AI debate
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
10
401
The man who coined "vibe coding" says he has never felt more behind as a programmer.
ALL I WANNA SAY IS THAT DON'T REALLY CARE ABOUT US
Andrej Karpathy:
"You can outsource your thinking
but you can't outsource your understanding"
In 30 minutes, you get Karpathy's whole playbook for working with agents.
Save it and watch it before you write your next line of code
Stanford now has a lecture on how Grok Bot works under the hood, taught by one of the people who built it.
SpaceXAI's Lee Robinson:
"Unlike a harness in your terminal, an always-on agent is more like a colleague
It knows when to be quiet"
Only tool names go into the prompt, long outputs go into files, screenshots go to sub-agents
In 45 minutes, you get the full architecture of an always-on agent, straight from a Stanford classroom.
Bookmark it now, this is the one you'll come back to
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
19
407
The people who built Grok Bot just opened up their own bot setup
Lauren Tan's engineering bots have food names
SpaceXAI's Lauren Tan & Peng Zheng:
"I mostly talk to my chief of staff, and my chief of staff talks to the eng lead
It lets me orchestrate really massive agent swarms. Sometimes I don't even look at the PR until after it's landed"
Worth more than any $2,000 AI course
Bookmark and watch it today
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
9
207
1.5M views in 30 days, and her social media team is Grok Bots
Elisa:
"I gave Grok Bot complete control of my social media
The team behind all of this is literally just me, my human assistant Zoe, and my Grok Bot team"
In 25 minutes, she walks through her whole setup: bots that find viral ideas, edit, clip, answer DMs, and what it costs vs hiring people.
Bookmark and watch it today
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
1
22
1,305
This is f*cking ALPHA
Greg Isenberg and Orgo co-founder Nick:
"Build your own custom Grok Bot for a business
And sell it for $25,000 upfront, then $3,000 a month"
In 50 minutes, they walk through the full stack, the pricing, and how to find your niche by calling 50 people.
Worth more than any $2,000 AI course
Bookmark and watch it today
12
267
This is f*cking insane
OpenAI CEO Sam Altman:
"In some narrow senses, like mathematical reasoning, we have super intelligence now
And I didn't think it was going to happen this fast"
In 49 minutes at Stanford, he explains why even he is rethinking how fast AI should go, and what it means for your job.
Worth more than any $4,000 AI course
Bookmark and watch it today
17
349








